Induction heating of moving metal strips: the optimal design of this devices by genetic algorithms and circuit models

A procedure for the optimal design of a transverse flux induction heating (TFIH) device is presented. An efficient genetic algorithm (GA), obtained by a modification of the classic genetic scheme, is introduced to direct the decisional process towards an optimization of main electrical and geometrical design parameters. The goodness of each proposed configuration is evaluated by solving, within the metal strip, the non-linear electromagnetic and thermal coupled problem. The solution of this three-dimensional (3D) time-varying problem is formulated by an integral approach, that is based on the subdivision of conductive regions in elementary volumes, in which uniform distributions of current density and temperature are assumed. Integrating Maxwell's equations in every elementary volume, eddy-currents and power losses are computed by means of an equivalent electrical lumped parameter network. By analogy, integrating the heat diffusion equation in the same elementary volumes, the temperature distribution is determined using an equivalent electrical lumped parameter network. Alternating the solution of electromagnetic and thermal equilibrium equations, associated to the equivalent electrical networks, the non-linear coupled problem is solved iteratively, in order to account for the temperature dependence of all physical parameters. The proposed procedure is applied to optimize the design of an experimental TFIH device, obtaining a maximum variation of the temperature distribution along the width of the moving metal sheet lesser than 7.5%.